The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documentation instructs the installation of an external dependency via npx skills add onsen-ai/chart-skill and requires running npm install for project dependencies. This involves executing code from a source (onsen-ai) that is not identified as a trusted vendor or associated with the skill author.
[DATA_EXFILTRATION]: The skill facilitates sending user-provided data to a remote third-party service at https://quickchart.io/chart using curl. This constitutes a network call to a non-whitelisted domain, which could result in data exposure if sensitive information is included in the chart datasets.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by ingesting untrusted data to generate chart visualizations across multiple engines. Ingestion points include data values provided to Vega, Vega-Lite, and ECharts templates. The skill's capabilities include subprocess execution, file system access, and network operations. While documentation recommends file-based inputs to mitigate shell injection, explicit boundary markers or sanitization logic are not present in the provided templates.
[COMMAND_EXECUTION]: The skill utilizes local shell commands to render charts, specifically node ${CLAUDE_SKILL_DIR}/scripts/render.mjs. The core rendering script (render.mjs) and several other mentioned scripts (such as generate.py and build_chart.py in the ECharts reference) are not included in the provided file list, preventing a complete security audit of the command execution logic.

nextreme-charts